branch: master
Commits on master
- 927f2dd wmma: add HIP FP16 to FP16 tensor core (#3287) 2 years ago
- 18e854c shrink MLB on sharded axis (#3255) 2 years ago
- a3652e6 minor cleanups to test_ops (#3290) 2 years ago
- 7725133 fix handcode_resnet50_opt.py (#3289) 2 years ago
- 9b8c1a0 Tensor.batchnorm works more than 2d and reuse in onnx (#3284) 2 years ago
- 7816c3b onnx update for trilu and argmax (#3283) 2 years ago
- 5b46b0f Simple RDNA3 emulator (#2974) 2 years ago
- 247a8a2 add canonicalization to View.create (#3280) 2 years ago
- d8f6280 hotfix: add CHECK_NEQ to fuzz_shapetracker_math 2 years ago
- 09f2952 reintroduce merge views in update benchmark (#3279) 2 years ago
- d298916 Revert "take merge views from corsix branch" (#3278) 2 years ago
- b57a16a take merge views from corsix branch (#3273) 2 years ago
- 6a4a5dc fix pad 0 size (#3277) 2 years ago
- b0a7552 cifar EVAL_BS set default value to BS (#3274) 2 years ago
- 861d5ac wmma: fix the upcasts after WMMA to be hcopt ordering invariant (#3250) 2 years ago
- af4ca85 MultiLazyBuffer.reshape new_axis without real_strides (#3272) 2 years ago
- 34c7621 HIP=1 NOCLANG=1 for tinybox external_model_benchmark (#3270) 2 years ago
- 085dc87 winograd should be 4 kernels (#3268) 2 years ago
- f48b6ac long running beam pool (#3267) 2 years ago
- 9e17378 Fix metal tests (#3266) 2 years ago
- 86748f4 fix bbox format to be a list (#3265) 2 years ago
- 67a7861 uoptimizer (#3262) 2 years ago
- 3ae811a tests for Tensor init data dtype and resulting dtype (#3247) 2 years ago
- 3c728d1 compiler support (#3260) 2 years ago
- 4273aab extra/gemm: add a simple_conv.py along with correctness check (#3236) 2 years ago
- 0aad8d2 rebuild ocelot (#3259) 2 years ago
- 4739351 use comgr to compile (#3248) 2 years ago
- c4d870d fix jit realize issue (#3258) 2 years ago
- 4197ef1 const cleanup with dtype.Scalar (#3257) 2 years ago
- 03a6bc5 move autogen to runtime/autogen (#3254) 2 years ago